Re: Site-wide vs. entry-specific protocol elements

[Robert Sayre <mint@xxxxxxxxxxxxxxx>]

I would love to take this approach if I thought it was workable. My 
reasoning follows, but I won't be surprised if most people agree with Paul.

Robert Sayre

Paul Hoffman / IMC wrote:
> Greetings again. The discussion on this list covers two different topics 
> that we might want to keep in separate documents. The things we're 
> talking about most now are protocol elements for sites (and we still 
> don't have a good definiton of sites), while the main atompub-protocol 
> document is specific to entries.

Entries with no author privileges, no way of setting permissions (aka 
draft status), no way of adding categories (something pretty much every 
system does), no way of managing uploaded content (something every 
system does), no way of commenting, no way of retrieving old entries 
when a new editor is pointed at a FeedURI (site management)...

> I propose that we *don't* add any site-wide protocol elements to the 
> current atompub-protocol document, but insteand create one or more 
> documents for the site-wide changes. There are a couple of advantages to 
> this:
>
> - Implementers who only want to help users add/edit/delete entries won't 
> have to deal with the other stuff

What are they adding to, where are the old ones, etc.? If they don't 
need to know that stuff, we can point them at the HTTP spec.

> - It will allow us to more likely meet our WG deadlines on the 
> entry-specific protocol document

There's a large body of work already done on site management in the form 
of RFC2518. We can avoid using the bad parts of it (the authors have 
helpfully written them down), and resist the temptation to invent our 
own stuff. We might finish quite quickly if we take advantage of the 
already-done products of other WGs.

> - It will give this group a bit more leeway in defining what a "site" is 
> without feeling rushed about the already-done entry stuff

The already-done entry stuff doesn't actually work right now. It needs 
the site management stuff to work. In particular, take a look at the 
FeedURI section of the protocol draft. I don't think editors will be 
real useful unless we firm that up. Even though I don't like 
PaceIntrospectionFile much, it is certainly an improvement on the status 
quo.

Robert Sayre